lp://staging/~jason-hobbs/maas/run-single-process-single-thread

Branch merges

Propose for merging

No branches dependent on this one.

Rejected for merging into lp://staging/~maas-committers/maas/trunk

Andres Rodriguez (community): Disapprove on 2014-10-19

Christian Reis (community): Needs Information on 2014-10-17

Diff: 12 lines (+4/-1)

1 file modified

contrib/maas-http.conf (+4/-1)

api

Related source package recipes

1 recipe using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Bug #1382575: Same node handed out in response to two parallel 'acquire' requests

Critical

Fix Released

Link a bug report

Related blueprints

3269. By Jason Hobbs on 2014-10-17

Only run a single wsgi process/thread for MAAS.

3268. By Andres Rodriguez on 2014-10-17

[r=rvb][bug=][author=andreserl] As per request from the release team, maas-proxy is now used on both, fresh install and upgrades.

3267. By Raphaël Badin on 2014-10-17

[r=rvb][bug=][author=rvb] Remove leftover XXX comment in the 1.7 release notes.

3266. By Julian Edwards on 2014-10-17

[r=rvb][bug=][author=rvb] Release notes for 1.7.

3265. By Graham Binns on 2014-10-17

[r=jtv][bug=][author=gmb] Add a Node.stop() method, a singular version of NodeManager.stop_nodes(). Functionally they're almost identical except that stop() only works on the current node.

This is further work towards fixing bug 1330765; there's a least one, probably two, branches to come after this.

3264. By Blake Rouse on 2014-10-17

[r=jason-hobbs][bug=1378527][author=blake-rouse] Fixes the images page to not show multiple resources for HWE enablement resources. Fixes the incorrect progress of downloading resources.

3263. By Newell Jensen on 2014-10-16

[r=blake-rouse][bug=1381213][author=newell-jensen] Setting default osystem and distro_series in mark_fixed. This fixes the issue for the performed steps (below):

1. Set the OS and Release to something other than the default.
2. Deploy a node.
3. Release the node, with wiping on.
4. Abort the wipe operation after a moment.
5. Mark node broken.
6. Mark node fixed.
7. OS/Release are not the default.

3262. By Gavin Panella on 2014-10-16

[r=gmb][bug=1381615][author=allenap] When starting monitors, cancel and reschedule an existing monitor with the same ID.

Previously, when a monitor was rescheduled, an existing monitor with the same ID was not cleaned up, and would eventually expire. Whichever of the monitors with that ID to expire or be cancelled first would work correctly. However, each monitor with that ID after that would crash with KeyError in monitor_expired().

3261. By Graham Binns on 2014-10-16

[r=jtv][bug=1330765][author=gmb] Add a start() method to the Node model.

This method is largely a distillation of what happens in bulk in NodeManager.start_nodes; most of the tests have been adapted from the tests for that method.

This is the first of several branches to move us away from batching starting and stopping nodes; Node.start() is not currently used anywhere and won't be until the other work is complete.

3260. By Blake Rouse on 2014-10-15

[r=rvb,newell-jensen][bug=1381796][author=blake-rouse] Fix get_effective_license_key in curtin_userdata_windows.